Jumping into ambitious frameworks like decentralised data architectures can backfire without sufficient organisational maturity. “You need to ensure your teams and domains are structured and ready,” as Aulona explains. A scalable, domain-oriented data strategy works best once teams operate independently with clear ownership, while earlier or smaller setups often benefit from starting centralised to deliver value faster. Organisational readiness extends beyond team structure to include scalable infrastructure, adequate technical literacy, and effective change management. Only when these foundations are in place can AI initiatives scale successfully, which in practice requires clear prioritisation and dedicated capacity across business, data, and engineering teams, treating AI as a strategic investment rather than a side project.

Innovation is important, but governance is what anchors a data strategy in trust and reliability, hence the need for strong governance policies to accompany the creative side of data innovation.

“Crafting a data strategy isn’t just about designing cutting-edge systems; it’s about ensuring those systems serve people in meaningful and compliant ways,” she explains. “A key takeaway for me has been that failing to prioritise governance from the start can lead to back and forth in the implementation and wasted time due to compliance issues.”

By balancing ambition with governance, organisations can drive sustainable growth without losing sight of accountability.

It is crucial to trigger a mindset shift across organisations to treat data as a product. This requires educating end users and empowering product owners to take full ownership of their data.

“Progress often stalls because product owners don’t fully understand their responsibility to govern and prioritise their own data effectively,” she notes. “It is a common mistake to only focus on the technical aspects, and forget that data strategy needs to be aligned with the business goals.

Otherwise in the end you you have all these advanced solutions, but it's just an additional cost if you're not really thinking with a product mindset: who is the end user of this data? Are they able to use it? Can they trust it?”

Flexibility is key: transformations should avoid rigid, comprehensive overhauls, instead favouring modular approaches that allow for evolution.

“A data lake can be used as a transitional layer, where organisations move step-by-step from centralised data usage to decentralised practices – or even machine-learning-ready solutions,” Aulona says.

This approach prevents large disruptions while enabling organisations to grow their maturity over time.

Data is only as useful as it is accessible, it is hence important to ensure teams create and share metadata for their datasets, making inter-organisational data discoverable. “Teams should build their own data repositories and document their metadata to facilitate internal collaboration,” she advises. A centralised data catalogue can then be developed to bring these individual contributions together, enabling streamlined access and fostering transparency.